Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232565287

A brief measure of children's behavior problems: The Behavior Rating Index for
Children.

Article  in  Measurement and Evaluation in Counseling and Development · July 1984

CITATIONS READS
15 4,529

1 author:

Arlene Rubin Stiffman


Washington University in St. Louis
76 PUBLICATIONS   2,960 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Arlene Rubin Stiffman on 17 May 2014.

The user has requested enhancement of the downloaded file.


A Brief Measure of Children's
Behavior Problems: The Behavior
Rating Index for Children
ARLENE R. STIFFMAN
JOHN G. ORME
DEBORAH A. EVANS
RONALD A. FELDMAN
PHOEBE A. KEENEY

This article reports the evaluation of the Behavior Rtlting t ndexfor Children, a l Svitem summated
category partition scale {hat provides a prorheuc measure of children's behavior problems,

Professionals who work with children are evaluation of children of differing ages or
continually confronted with a need for of the same child at different ages. In all
short, reliable. easy-to-administer mea- these situations, the evaluation of a child's
sures that apply to a broad range of issues. behavior must be valid and reliable.
It is frequently critical to have informa- This article describes the development
tion about children's behavior from in- and evaluation of the Behavior Rating I ndex
dividuals who observe them in a variety [07- Children (BRIC), which is capable of
of environments. Potential respondents being used in all the above evaluative sit-
include parents. teachers, therapists, group uations. The BRIe is essentially a pro-
leaders, and the children themselves. Fur- thetic measure of childhood behavior. It
thermore. many research evaluations, measures only the degree of a behavior
particularly those in school and comrnu- problem. as opposed to metathetic instru-
nity settings, demand data about the same ments that measure the direction or type
child over a long period, including im- of such problems (Hudson, 1982).
mediately before and after treatment and Although numerous and various ques-
at longitudinal posttreatment intervals. tionnaires purport to measure children's
Research of this type often requires the behavior problems (Arnold & Smeltzer
1974; Borgaua & Fanshel 1970; National
Arlene R. Stiffma.n is Co-Din'ctoT, Center for Ado- Institute of Mental Health. 1980; Quay,
lescent Mental. Health. Utashington University. St. 1977; Spivak. Swift. & Prewitt, 1971), none
Louis. John. Go Orme is a postdoctoral fcllo<!\ D4.'- can be used by multiple respondents for
partmeut of Behavioral Sciences, L'nil.'enitJ or Chi- children of all ages. Although many of the
cago_ Deborali A. Euans is a system.s analyst, St. Louis existing measures possess adequate reli-
County Governmenl. Ronald .1.. Feldman is Director ..
ability and validity, practical problems limit
Center for Adolestrru .Heutaf Health, JrasJu°ngton
their use for field or clinical research re-
University. Phoebe it [(tent?) is a research assistant,
garding the amelioration or prevention of
GAlN Program, Washillgton Universit», The prep-
aration of this article has been [unded b)' research behavior problems in children. Some of
s=« MH350;3 fro",lhe National Institute oj Mental. the available instruments are appropriate
Health. Office of Prevention. The authors graJefully only for children of restricted age ranges;
acknowledge the sUppo>t of the NJlI,fH. thus they cannot be used for longitudinal

JULY 1984

83
,
I

studies or studies of children who differ discern evidence of extreme responses on


greatly in their respective ages. Most ques- the part of respondents, Each item is rated
tionnaires cannot accommodate the col- according to a 5-point Liken-type system
lection of comparable information from (l = rarely or never: 2 = a little of the
multiple respondents-children. parents, time; 3 = some of the lime; 4 = a good
teachers, and therapists. Dissimilar instru- part of the time; 5 = most or all of the
menrs make it difficult to compare be- time).
havior from one setting to another. The behavioral problems that were se-
Many brief instruments actually require lected for inclusion in the BRIC represent
15-30 minutes to complete. Although this behavioral dimensions or factors that ap-
length is short for detailed behavior pro- pear repeatedly in established behavioral
files. it is unwieldy for frequent test-retest inventories (Achenbach. 1978-1979;
evaluations, is highly intrusive in treat- Achenbach & Edelbrock, 1978}, in the
ment situations, and is time consuming DSM-III Classification System for Chil-
for evaluations of groups or classes of chil- dren (American Psychiatric Association,
dren (a particularly critical issue for school 1980), and in other instruments, regard-
and community evaluations). Conse- less of age and sex. The BRIC, however,
quently, there is a need for a brief, ac- was designed simply to measure the de-
curate instrument that is appropriate for gree of problem behavior.
evaluating children of widely differing The ten items that denote problem be-
ages, that is available in comparable form haviors are used to compute a summary
for multiple informants, and that. applies score for the BRIC. The total score for
to group and classroom settings. the BRIC is computed by employing a for-
The authors of this article needed such mula proposed by Hudson (1982):
an instrument for a program of preven- S = (Y - lvi (100) -7 4N. This formula
tive mental health research. A review of provides a simple linear transformation
the available literature revealed no instru- that results in a BRIC summary score that
ments that met all the above criteria. Con- has a potential range of 0-1 00, regardless
sequently, the authors developed the BRIC of the number of items completed. Higher
in a form for a respondent who is eval- scores indicate more severe behavioral
uating a single individual and in a form problems.
for a respondent who is evaluating a group
or classroom of children (Figure 1). METHOD

THE BEHAVIOR RATING The BRIC was evaluated through its use
INDEX FOR CHILDREN with more than 600 children who partici-
pated in a field experiment known as the
The BRIC is a 13-item. summated cate- GAIN (Group Activities for Individual
gory partition scale (Stevens, 19681. Of Needs) Program. The participants included
the 13 items, 3 randomly placed items both referred and nonreferred children.
(items 1, 6, and 10) indicate a behavioral The nonreferred children were regularly
strength of the respondent. They are not enrolled members of the Jewish Commu-
calculated in the behavior problem score; nity Center Association. The referred chil-
only the 10 problem-oriented items are dren were recruited as participants because
scored. The 3 strength-oriented items are one or both of their parents were in treat-
embedded in the BRIe so that the re- ment for a clinically diagnosed mental ill-
spondents are not confronted solely with ness. The available research literature clearly
a list of "undesirable" behaviors and to demonstrates that their parents' illness places

MEASUREMENT AND EVALUATION IN COUNSELING AND DEVELOPMENT


\

84
o

Group name Dete


Re.pondent
PI••• plata the most appropriate numoer in the bOK beside each child's name. The number refen to the frequency with which a given problem is
•• hlbited,
1 • rarelv or never
2 - • IIttlo of the time
3 • tome of the time
4 Q good part of the time
;10

6 • m05t or en of the time

--------
Child', neme
.. ._- I,
f--- ~-
2, 3, 4, 6, 6, 7,
I 8, 9, 10, 11, 12. 13. 14,
--f--
1. app",,, to be relaxed or hoppy

2. hide. hi. thoughts from oth.,.

3. says or does reol!y strange things

4. doe. not pay attentlcn when he should


-
- . ...
5. quits a job or t8ik without finis:hing it

6. gets along well with otho"


-
7, hits, puih8l. or huru others
-
8, gots along poorly with others

9, go•• very upset


-
---
10. compliments or helps others
- ----.
11. ,•• I. sick
.. ----_ .
12. cheat.

c: 13. I•••• temper


C
r-
-
-<
FIGURE 1

..
'"
0> The Behavior Rating Index for Childre" (BRie) Group Report Form (CoPVright © Stiffman. 1983)

,j
them at high risk for behavioral problems Response Set and Bias
(Anthony, 1978; Clausen & Huffine, 1979;
Rutter, 1966). Neither response set nor nonresponse to
The children participated in weekly ac- individual items presented a problem. For
tivity groups that attempted to enhance only 6 of the 875 completed forms did
their social competencies, coping skills. and respondents complete fewer than eight of
task abilities. Over 60 groups of children the ten scored items. Only one respondent .
(each consisting of a group leader and ap- completed both the positivity and nega-
proximately eight children) participated tivity worded items in the same extreme
during the 2 years of the program. fashion.
During the program, the BRIC was Validity
completed by the participant children at
repeated intervals (four times a year). The The BRIe's concurrent validity and con-
referred children's behavior also was eval- struct validity are adequate. ConcuTTenJ va-
uated both before and after group treat- lidity refers to the degree of correspondence
ment by their parents or guardians (n between a given measure and external cri-
= 183), by their teachers (n = 198), by teria (Cronbach, 1971). It was found that
group leaders (n = 28), and by nonpar- the phi correlation between referred chil-
ticipant observers (n = 13). The resultant dren's scores on the BRIC, as reported by
scores then provided data to determine parents, and children's treatment status,
the internal consistency reliability of the as reported by parents, was .65 (p < .001).
BRIC. Because not all the children with behav-
At the same limes they completed the ioral problems were treated for such
BRIC, the parents and teachers of re- problems and because all the children who
ferred children (n = 306) also completed were in treatment did not necessarily dis-
the lengthier (and thus unwieldy for group play problems (some were in family ther-
measurement) lI8-item Child Behavior apy because of parental problems), a
Checklist (CBC) (Achenbach, 1978-1979). perfect correspondence between these two
This permitted the assessment of concur- variables was neither expected nor found.
rent validity. In addition, a sample (n = 24) Construct validity was assessed through
of the referred parents and children com- the relationship between the parents' re-
pleted the entire intake interview on a 2- port of the referred children's behavior
week test-retest basis. This provided data on the BRIC and the parents' report of
for assessing reliability by means of test- this behavior on the CBC. The BRIe scores.
retest correlations and intraelass correla- from parents demonstrated high positive
tion coefficients. Furthermore, approxi- correlations with the CBC behavior prob-
mately half of the leaders and observers lem scores (r = .76, P < .001). This cor-
completed the BRIC twice within I week relation is strong, especially because a small
to provide data for determining test-retest number of test items tends to attenuate
reliability coefficients and interrater cor- reliability.
relations for their forms. Internal Consistency

RESULTS Internal consistency reliability was ex-


amined through coefficient alpha (Cron-
Table I shows the response bias, validity. bach, 1951), a measure that is based on
internal consistency, repeated-measures all of the interirem correlations within an
reliability, and inrerrater reliability that instrument. Coefficient alpha was com-
were evaluated. puted separately for each group of re-

MEASUREMENT AND EVALUATION IN COUNSELING AND DEVELOPMENT

86
./

TABLE 1
Validities and Reliabilities of the BRIC for Parents, Teachers,
Group Leaders, Observers, MId Children

Validity Reliabili~.
Treatment Internal Test..•_
__
BRie nt stetus cac consist8ncv Crand Interreter
(Phil Ir) (Alpha) intracl_1 Ir)
Parents .65" .76" .81·· .71-.72" .20 with
children·
Teachers na na .86·· na na
Groups leaders n8 na .86*· .89-.92·· .51 with
observers"
Observers n8 na .so" .84-.89--
Children na na .60-.70" .50"

*p < .001.
•'p <.0001.
na = not assessed.

spondents. It was consistently higher for week baseline period and at the end of
the BRICs completed by adults than for treatment (approximately 25 weeks later).
those completed by children. Alpha was To obtain test-retest data on parents,
.81 for parents, .81 for group leaders, .80 24 referred families returned within 1 or
for observers, and .86 for teachers. In 2 weeks of the initial interview for a sec-
contrast, it was .60 for nonreferred chil- ond interview. These families were re-
dren and. 70 for referred children. cruited from the second year's cohort of
Therefore, the internal consistency reli- referred families. All referred families were
ability of the BRICs that were completed asked to participate. The first 24 families
by children seems to be marginal while who agreed to be reinterviewed com-
the reliability of the BRICs completed by prised the sample. At the second inter-
adults is adequate (Guilford & Fruchter,
view, the parents completed a second
1978; Hudson, 1982).
BRIC. Test-retest data for the BRICs
Repeated-Measures Reliability completed by children were not collected
systematically. Because the referred chil-
Repeated-measures reliability is fre-
dren completed the BRIC at intake and
quently considered critical to any pro-
at the end of baseline, these retests formed
gram that attempts to measure change
the nexus of test-retest data for children's
over time (Kerlinger, 1973). Each group
BRICs.
of respondents (except teachers) that
completed the BRIC was asked to do so Repeated-measures reliability from all
again on a test-retest basis. Group lead- respondents consistently demonstrated a
ers (n = 1I) and observers (n = 6) from level that indicates the BRIC is an appro-
the second year's cohort participated in priate measure of behavioral change. Test-
two test-retest periods. They were asked retest reliability for BRICs completed by
to rate the children in their groups (n group leaders was .89 (J1 < .001) after they
= 155 and n = 84, respectively) twice had only 4 hours of contact with the chil-
within 1 week, both at the end of a 4- dren. (The groups had met four times for

JULY 1984

87
I hour each time. Not all children at- Interrater Reliability
tended every session.)
The intraclass correlation coefficient, a Interrater reliability is computed to de-
more stringent measure of repeated-mea- termine whether two different respon-
sures reliability because it controls for di- dents, observing the same behavior, report
rectional drift (Bartko, 1976), was also .89. similar scores on an instrument. Observ-
Test-retest reliabilities for BRICs corn- ers and group leaders and parents and
pleted by observers was .84 after less than children simultaneously completed their
1 hour of total contact with the children versions of the BRIC.
(four observation sessions of 15 to 20 min- Interrater reliabilities between BRICs
utes each). The intraclass correlation coef- completed by observers and group leaders
ficient for observers was .85 <.p < .000 I). and those completed by parents and chil-
These repeated-measures reliabilities for dren were not nearly as high as the test-
group leaders and observers, which were retest or internal consistency reliabiliries. In
measured immediately after baseline, in- p.-.1l"t,
this is to be expected; different re-
dicate that with minimal observation, the spondenrs typically do not see children un-
BRIC yields reliable data. When the same der exactly the same circumstances, nor do
test-retest procedure for observers and they evaluate behavior according to exactly
group leaders was completed at postlest, the same standards. The correlation be-
the intradass correlation coefficients and tween BRIC scores for group leaders and
Pearson rs reached .92 for group leaders nonparticipant observers was only .51
and .89 for observers. Cp < .001, n = 84). Such interrater relia-
The repeated parent interview dem- bility seems adequate, especially when one
onstrated that the lest-retest reliability for considers the brevity of the instrument and
BRICs completed by parents was. 72 the difference between the two respon-
(P < .0001). The intraclass correlation dents in the form of contact and the length
coefficient was .71. Considering that the of contact with the children. It certainly
test-retest reliability of the much longer compares well with interrater reliability for
CBC ranges from .82 to .90 (Achenbach, oilier instruments. The CBC, which is al-
1978-19i9), .71 seems to be adequate for most ten limes longer, yields an interparent
such a short instrument. correlation ranging only from .54 to .74
Unfortunately, systematic test-retest (Achenbach, 1978-1979).
comparisons of BRICs completed by Parents and children completed the BRlC
children were not obtained. Test-retest at the same time. It was expected that there
reliability computed between intake and would not be extensive agreement between
baseline for the referred subjects (n the behavioral ratings by parents and chil-
'" 180) and controlled for the time in- dren; that expectation held true. The cor-
terval was .50 (P < .0001). It should be relation between parent and child reports
noted that the testing situations differed of the BRIC is only .20 (p < .001) for chil-
markedly: one was a personal interview dren ages 6 to 15 (n = 306). Although it
and the other an anonymous group test- changes to .28 (p < .001) when only chil-
ing situation. Furthermore, the reliabil- dren over 12-years old are included, it is
ity did rise to .58 for those children who certainly low. This suggests that the chil-
were 12 years old or older, indicating dren see their behavior quite differently than
that older children tend to be more re- do their parents.
liable reponers. It is clear that BRICs Because none of the other respondents
completed by children are less reliable completed the BRIC at the same time or
than those completed by adults. were evaluating behavior in similar situ-

MEASUREMENT AND EVALUAnON IN COUNSEUNG AND DEVELOPMENT

B8
[

oflarge groups of children prior to more Borgaua. E.F .. & Fanshel, D. (J 970). The Child
extensive metathetic testing of those chil- Behavior Characteristics (CBe) form: Re-
dren who are evidencing problems. vised age-specific forms. A:fullivm7ate Be-
With the exception of the children's iunsiorol Research, 5, 49-82.
Clausen, J.A .. & Huffine. c,L. (19791. The im-
form, the BRIe generally demonstrates
pact ot parental mental illness on children.
satisfactory reliability and validity for di-
Research ill Commuuitv Mental Health, J.
verse groups of respondents who are fa- 183-214. -
miliar with a child's behavior. This Cronbach, LJ. (195 I). Coefficient alpha and
instrument should be a valuable addition the internal structure of tests. Psscheme-
to the diagnostic arsenals of clinicians, trika. 16. 297-334.
teachers, group leaders, and researchers Cronbach, LJ (197Ii. Tesl validation. In R.L
who seek an instrument that possesses not Thorndike (Ed.), Educational measurement
only satisfactory psychometric character- (2nd ed.) (pp , 443-507). Washington:
American Council on Education.
istics but also offers case of administration
Guilford, ].P .. & Fruchter, B. (1978). Funda-
and appropriateness for a wide range of
menial suuistics in psycholugy and eduauion (6th
individuals.
ed.). New York: McGraw-Hill.
Hudson, W.W. (1982). The clinical measurement
packnge: it field manual. Homewood, IL:
REFERENCES Dorsey.
Kerlinger, F.N. (1973). Foundations of behaviora!
Achenbach, TM. (1978-1979). The Child Be- "search (2nd ed.). ;";e"· York: Holt, Rine-
havior Profile: An empirically based sys- hart and Winsron.
re m for assessing children's behavior National Institute of Mental Health. Series AN
problems and competencies. International 1'10. I. (1980). 71!e assessment of ps),chnpath-
[ournal of Menial Health, 7, 24--42. ology muJ behaoioral "rob/e-ms in children: A
Achenbach. T.M., & Edelbrock, C.S. (978). review rf scales suitable for epidemiological ami
The classification of child psychopathol- clinical research (l96'7-J979i (DHHS Pub-
ogy: A review and analysis of empirical lication No. Am,1 80-H);l7). Washington,
efforts. 85, 1275-130 I.
P!»)'choftJ~urtl Bulletin. DC: Government Printing Office.
American Psychiatric Association. (l980). Di- Quay, H.C. n 977). Measuring dimensions of
agnostic mid statistical ma-nual uf mental dis- deviant behavior: The Behavior Problem
orders (Srd ed.). Washington, DC: Author. Index. [oumal of Abnormal Child Psyclwlogy.
Anthony, E..J. (1978). From birth to break- 5,277-287.
down: A prospective studv or vulnerabil- Rutter. M. (1966). Children of sick parents: .4n
ity, In EJ Anthony, C. Koupernik, & C. enoironmentol and psychiatric siud». London:
Chiland (Eds.), The child ill his family: V!ll- Maudsley Monograph 16.
nerablechiidren (Vol. 4) (pp, 273-286). New Spivak, G., Swift, G., & Prewitt, J- (1971). Syn-
York: Wile,·. dromes of disturbed classroom behavior:
Arnold. E.. & Smeltzer. D.]. (197-iJ. Behavior A behavioral diagnostic system for ele-
checklist factor analysis for children and rnenrarv schools. Journal of Special Educa-
adolescents .. +rchives of Genera! Vl.w:hialry. lion, 5. 269-292.
30. 799-804. Stevens, SS (1968). Ratio scales of opinion. In
Bartko. J.J. (1976). Oil various intraclass cor- D.K. Whitla (Ed.). Hmldbook of measurement
relation reliability coefficients. Psvchological and as..us..
~lJll.·111 in behninorai scinta (PI)' 171-

Bulletin, 83. 762-765. 199). Reading, MA: Addison-Wesley.

MEASUREMENT AND EVALUATION IN COUNSELING AND DEVELOPMENT

90
ations, the interrater correlations between "glioses when there is no need. More ex-
BRICs completed by parent and teacher. tensive and time-consuming testing could
by teacher and group leader. by teacher he reserved for those children who seem
and observer. and by parent and group to warrant it on the basis of speedier pre-
leader were not computed. Children's be- liminary screening devices. such as the
havior is known to be situation specific. BRIe.

Clinical Cutting Point


DISCUSSION
Frequently. clinicians or researchers who
use a brief prothetic screening device such The BRIe offers several practical advan-
as the BRIC require some form of clinical tages for clinicians and researchers. It is
cutting point by which they can evaluate extremely brief and can be completed in
not only the extent of a problem, but less than 3 minutes by adults or children.
whether or not the problem seems severe Therefore. it can be administered easily
enough to warrant further testing. diag- at repeated intervals over the course of
nosis. or treatment. There are several treatment and at follow-up intervals with-
methods for developing such a cutting out intruding unduly on treatment time.
point. All involve the use of established It is especially appropriate for group test-
measures. which can indicate whether or ing situations in which a large number of
not a child has a clinically significant be- children need to be evaluated expedi-
havior disorder. Because the CBC score tiously, Also. it can be scored quickly and
was available from parents' data. cumu- easily by hand. Furthermore. BRIe scores
lative frequencies were computed for both range from 0 to 100. regardless of the
the referred children's CBC and BRIC number of items completed. Thus, the
scores to determine a cutting point for the range of possible scores is standardized
adult versions of the BRIC. and readily comparable. The BRIC can
Using the CBC's clinical cutting point be used in parallel forms by many differ-
of63 as a criterion. parent-reported BRICs ent respondents, including parents,
demonstrated a cutting point of 30. Only teachers. therapists, and children. Finally,
16% of the children falling within the CBC the BRIC is appropriate for both male
clinical range had BRIC scores lower than and female children who are 7- to IS-years
30. Thus, using scores greater than 30 as old.
indicative of a clinical problem results in Analyses of response set. validity, reli-
a misclassification rate of only 18%. This ability. and clinical cutting points dem-
is well within the misclassification range onstrate that the BRIC is a promising new
for other longer instruments. instrument that possesses appropriate
Under no circumstances should a cli- psychometric and practical characteristics
nician or researcher automatically assume for practice and research. This is espe-
that children with scores above 30 have cially [rue for the BRIe forms that are
clinical problems and those under 30 do used by adults. Response set poses no
not. Scores above 30 can be considered to problems. Internal consistency reliability
indicate that more detailed metathetic ~s adequate. Interrater reliability is mar-
testing devices. which are designed to de- ginal but adequate. Test-retest reliability
termine specific diagnoses. should be used. is high. Concurrent and construct validi-
Thus the BRIe could effectively serve as ties are adequate. Furthermore. a clinical
a preliminary screening device for large cutting point of 30 results in an acceptable
numbers of children. Its use could vitiate misc1assification rate, so the instrument
the need for unnecessarily detailed di- seems appropriate for prothetic screening

JULY 1984

89

View publication stats

You might also like